INTRODUCTION
Type 2 diabetes (T2D) is characterised by insulin resistance and progressive deterioration of β-cell function, ultimately result-ing in the absolute deficiency of insulin secretion that is mani-fested by non-response to escalating oral hypoglycemic agents
(OHAs) to overcome hyperglycemia.1-3 As proof-of-concept,
earlier and more intensive insulin initiation in patients with newly diagnosed T2D resulted in not only a great improve-ment in glycemic control, but also a modified natural history of diabetes through restoration and preservation of pancreatic β-cell function.4,5 In this regard, some controversy might remain
over whether appropriate implementation of insulin ana-logues, even in subjects with a moderate duration of T2D, is ex-pected to achieve the targeted HbA1c while minimising side effects such as hypoglycaemia.6,7 To address this, many
ques-tions should be clarified in real clinical practice, such as when to initiate insulin (before or after islet failure), which patients to initiate treatment in, which kind of insulin analogue (basal, rapid-acting, biphasic, or combination of basal and rapid-act-ing insulin analogue) should be used, and how to administer insulin (starting dose, number of injections, divided ratio of to-tal doses, and titration methods).8-10 Previously, we
investigat-Characteristics Predictive for a Successful Switch
from Insulin Analogue Therapy to Oral Hypoglycemic
Agents in Patients with Type 2 Diabetes
Gyuri Kim
1,2*, Yong-ho Lee
1,2*, Eun Seok Kang
1,2, Bong-Soo Cha
1,2, Hyun Chul Lee
1,2, and Byung-Wan Lee
1,21Department of Internal Medicine, Yonsei University College of Medicine, Seoul; 2Graduate School, Yonsei University College of Medicine, Seoul, Korea.
Purpose: The objective of this study was to investigate clinical and laboratory parameters that could predict which patients could maintain adequate glycemic control after switching from initial insulin therapy to oral hypoglycemic agents (OHAs) among patients with type 2 diabetes (T2D).
Materials and Methods: We recruited 275 patients with T2D who had been registered in 3 cohorts of initiated insulin therapy and followed up for 33 months. The participants were divided into 2 groups according to whether they switched from insulin to OHAs (Group I) or not (Group II), and Group I was further classified into 2 sub-groups: maintenance on OHAs (Group IA) or resumption of insulin (Group IB).
Results: Of 275 patients with insulin initiation, 63% switched to OHAs (Group I) and 37% continued insulin (Group II). Of these, 44% were in Group IA and 19% in Group IB. The lowest tertile of baseline postprandial C-peptide-to-glucose ratio (PCGR), higher insulin dose at switching to OHAs, and higher HbA1c level at 6 months after switching to OHAs were all associated with OHA fail-ure (Group IB; p=0.001, 0.046, and 0.014, respectively). The lowest tertile of PCGR was associated with ultimate use of insulin (Group IB and Group II; p=0.029).
Conclusion: Higher baseline level of PCGR and lower HbA1c levels at 6 months after switching to OHAs may be strong predictors for the successful maintenance of OHAs after switching from insulin therapy in Korean patients with T2D.
Key Words: Hypoglycemic agents, insulin, type 2 diabetes mellitus Yonsei Med J 2016 Nov;57(6):1395-1403
http://dx.doi.org/10.3349/ymj.2016.57.6.1395 pISSN: 0513-5796 · eISSN: 1976-2437
Received: January 6, 2016 Revised: March 9, 2016 Accepted: March 14, 2016
Corresponding author: Dr. Byung-Wan Lee, Department of Internal Medicine, Yonsei University College of Medicine, 50-1 Yonsei-ro, Seodaemun-gu, Seoul 03722, Korea.
Tel: 82-2-2228-1938, Fax: 82-2-393-6884, E-mail: bwanlee@yuhs.ac *Gyuri Kim and Yong-ho Lee contributed equally to this work. •The authors have no financial conflicts of interest.
© Copyright: Yonsei University College of Medicine 2016
ed the stage at which both physicians and patients agreed to start insulin therapy, and what are acceptable HbA1c levels for Korean subjects with T2D in real practice.11 We have also tried
to answer questions regarding the characteristics of patients who respond adequately to basal plus insulin or to insulin an-alogues (basal or biphasic insulin analogue) after sulfonylurea failure,12,13 as well as the optimal dosing ratio of twice-daily
bi-phasic insulin analogues.14,15 However, little research has been
performed on whether a change in therapeutic strategy from insulin therapy to OHAs could be successfully applied and, if so, which characteristics are predictive of successful mainte-nance on OHAs after switching from initial insulin therapy in patients with T2D, especially in the Korean population.
Therefore, in this study we aimed to investigate the clinical and metabolic parameters of T2D patients who could main-tain adequate glycemic control after switching to OHAs from insulin analogues.
MATERIALS AND METHODS
Study population and study design
In this longitudinal retrospective study, we analyzed subjects with T2D from three cohorts who had enrolled in previous stud-ies. These patients were randomly assigned to a long-acting in-sulin analogue and biphasic inin-sulin analogue, or were switched from OHAs to twice-daily of premixed insulin lispro 25 or pre-mixed insulin aspart protamine 30.12,14,15 All patients were
fol-lowed up in October 2014 at Severance Diabetes Center, Seoul, South Korea. Patients were excluded for the following reasons: less than 6 months of follow-up after the recruited study end-point, secondary disorders of hyperglycemia, such as iatrogenic, inflammatory, or endocrine disorders during the observation period.
We reviewed and rechecked electronic medical records to assess whether the subjects who stopped taking insulin later
resumed insulin or continued on OHAs after insulin discontin-uation. The study subjects were first classified into two groups according to whether they switched from insulin to OHAs: the off-insulin group (Group I) and the on-insulin group (Group II) (Fig. 1). Because of the nature of real-world and retrospective study, there was no study-related guideline or consensus on the selection of patients and the choice of OHAs but fully at the discretion of the physician based on clinical judgments and the levels of HbA1c at switching to OHAs from insulin were approximately less than 7.5%. If subjects were able to discon-tinue insulin treatment at least once, they were classified into the off-insulin group (Group I). The off-insulin group was fur-ther divided according to resumption of insulin during the follow-up period into the off-insulin with maintained OHAs (Group IA) and off-insulin with re-initiation of insulin (Group IB). Sulfonylurea failure was defined as either glimepiride >4 mg, gliclazide >120 mg, or glibenclamide >10 mg per a day. Symptomatic hypoglycemic events were defined as blood glu-cose below 70 mg/dL concomitant with symptoms such as sweating or palpitation. The protocol of this study, adhered to the tenets of the Declaration of Helsinki, was approved by the Institutional Review Board (IRB No. 4-2014-0059) of Severance Hospital and written informed consent for this study was not required by the institutional review board because the research-ers only accessed the database for analysis purposes and per-sonal information was not used.
Clinical and laboratory measurements
Body mass index (BMI) was defined as weight divided by the square of the height (kg/m2). Measurement techniques
includ-ed the hexokinase method for both fasting and postprandial glucoses and the enzymatic colorimetric method for total cho-lesterol, triglycerides, and high-density lipoprotein cholesterol. HbA1c was measured by high-performance liquid chroma-tography using VariantTM II Turbo (Bio-Rad Laboratories,
Her-cules, CA, USA). Serum glycated albumin (GA) was determined by an enzymatic method as previously described.16 The
[image:2.595.42.282.537.707.2]refer-ence ranges for HbA1c was 4.0% to 6.0%, and for GA was 11.0% to 16.0%. Levels of serum C-peptide were measured in dupli-cate using an immunoradiometric assay method (Beckman Coulter, Fullerton, CA, USA). Pancreatic β-cell functions in the context of ambient insulin resistance were assessed using the following indices: 1) fasting or postprandial C-peptide; 2) fast-ing glucose ratio or postprandial C-peptide-to-glucose ratio (PCGR) [fasting or postprandial C-peptide level (ng/mL)/fasting or postprandial glucose level (mg/dL)×100]; 3) index for C-peptide (postprandial C-peptide-fasting C-pep-tide)/(postprandial glucose-fasting glucose); 4) homeostasis model assessment of b-cell function [fasting insulin (μIU/ mL)×20/fasting glucose (mmol/L)-3.5]; and 5) homeostasis model assessment of insulin resistance [fasting insulin (μIU/ mL)×fasting plasma glucose (mmol/L)/22.5]; 6) the fasting or postprandial secretory units of islets in transplantation (SUIT) Fig. 1. Changes in regimen of subjects with type 2 diabetes. OHAs, oral
hypoglycemic agents.
Ultimately treated with insulin (n=153, 56%) Successfully continued OHAs
(n=122, 44%)
Group II Group I
Group IA
Group IB Group IB and II
Initiation of insulin therapy (n=275)
Changed to OHAs
(n=174, 63%) Re-changed to insulin
therapy (n=52, 19%)
index {[250×fasting or postprandial C-peptide (nM)]/[fasting or postprandial blood glucose (mM)-3.43]}.17-19 A
standard-ized mixed meal (Ensure, Meiji Dairies Corporation, Tokyo, Ja-pan; total 500 kcal, 17.5 g fat, 68.5 g carbohydrate, and 17.5 g protein) stimulation test was conducted after an 8-hour over-night fast, and postprandial glucose and C-peptide were mea-sured at 90 minutes in 33.1% of subjects. In the rest of 66.9% of subjects, postprandial glucose and serum were measured 2 hours after each subject’s conventional diet.12,14,15
Statistical analysis
All continuous variables are presented as mean±standard devia-tion (SD) and categorical variables are presented as propordevia-tions. Data are shown as mean±standard error (SE) in the figures.
Differences were analysed using Student’s t-test for continu-ous variables and chi-squared test for categorical variables. Cox regression analysis was performed to assess whether
[image:3.595.59.554.317.672.2]mainte-nance of OHAs or resumption of insulin treatment after insu-lin discontinuation was associated with various cinsu-linical and laboratory parameters at baseline (Model 1) and during the follow-up period (Model 2). Multiple linear regression analy-sis was performed to determine the independent relationship of the studied variables. The odds ratios (ORs) and 95% confi-dence intervals (CIs) for the factors associated with subjects who are ultimately unable to maintain the use of OHAs with-out insulin were calculated using multivariate logistic repres-sion analysis. Cumulative event rates for resumption of insulin were estimated by Kaplan-Meier survival curves, and proba-bility values were calculated with the log-rank test. The receiver operating characteristic (ROC) curve of PCGR was used to as-sess the optimal cut-off value for the prediction in successful maintenance of OHAs with the areas under the curve (AUC) with maximum Youden index (sensitivity+specificity-1). A p value <0.05 was considered statistically significant. Statistical
Table 1. Baseline Characteristics of Subjects
Group I
Group II (n=101) Group IB and II (n=153) Group IA (n=122) Group IB (n=52)
Age (yr) 59.1±10.8 58.4±10.9 57.9±11.1 58.2±11.0 Sex, men/women (% men) 66/56 (54) 33/19 (63) 68/33 (67) 101/52 (66)ǁ
BMI (kg/m2) 25.1±3.4 24.6±2.8 24.0±3.1† 24.2±3.0ǁ
Obesity (%) 63 (52) 24 (46) 37 (37) 61 (40)ǁ
Duration of diabetes (yr) 8.7±6.2 8.9±6.4 10.0±7.1 9.6±6.9 Family history of diabetes (%) 50 (48) 15 (33) 40 (46) 55 (36) Fasting plasma glucose (mg/dL) 173±63 191±85 177±68 180±74 Postprandial glucose (mg/dL) 274±81 303±83‡ 294±95 297±90
Fasting C-peptide (ng/mL) 2.43±1.10 2.40±2.35 2.39±1.70 2.40±1.95 Postprandial C-peptide (ng/mL) 4.83±2.05 4.49±3.31 4.68±3.56 4.61±3.46
HbA1c (%) 9.1 9.7‡ 9.3 9.5
Glycated albumin (%) 25.5±8.0 28.3±10.8 26.3±9.4 27.0±9.9 Creatinine (mg/dL) 1.03±0.84 0.98±0.41 1.08±0.70 1.05±0.62 Total cholesterol (mg/dL) 162±39 174±42 158±39 162±42 Triglycerides (mg/dL) 168±88 142±80 177±239 168±204 HDL cholesterol (mg/dL) 42±12 42±12 42±12 42±12 LDL cholesterol (mg/dL) 84.9±30.9 96.5±38.6‡ 84.9±30.9 88.8±34.7
FCGR 1.59±1.03 1.52±1.49 1.57±1.56 1.55±1.53 PCGR 1.94±0.92 1.59±1.33 1.66±0.84 1.63±1.03§
Lowest tertile (%)* 24 51§ 36 41¶
Higher tertiles (%)* 76 49§ 64 59¶
ICI 0.024±0.034 0.025±0.059 0.006±0.095† 0.013±0.084
HOMA-β 61.29±142.24 54.43±89.84 86.59±182.13 75.39±156.66 HOMA-IR 6.43±17.08 4.64±8.00 6.21±9.61 5.66±9.08 Fasting SUIT index 46.2±52.5 44.9±54.0 49.1±72.9 47.6±66.7 Postprandial SUIT index 39.6±21.7 31.5±27.6 33.4±18.1 32.7±21.8ǁ
BMI, body mass index; HDL, high density lipoprotein; LDL, low density lipoprotein; FCGR, fasting glucose ratio; PCGR, postprandial C-peptide-to-glucose ratio; ICI, C-peptide-genic index; HOMA-β, homeostasis model assessment of b-cell function; HOMA-IR, homeostasis model assessment of insulin resis-tance; SUIT, the secretory units of islets in transplantation.
Data are expressed as mean±standard deviation.
*Chi-squared test, †p value <0.05, Group I vs. Group II, ‡p value <0.05, Group IA vs. Group IB, §p value <0.001, Group IA vs. Group IB, ǁp value <0.05, Group IA vs.
analyses were performed using PASW Statistics version 20.0 for Windows (SPSS Inc., Chicago, IL, USA).
RESULTS
Patient enrolment and classification
Of 363 T2D participants who had enrolled in three previous studies, 305 insulin-naïve subjects (84%) were well document-ed and available for clinical follow-up up to October 2014. Af-ter exclusion of subjects with sAf-teroid use (n=15), and less than
6 months of follow-up period after study enrolment (n=15), a total of 275 patients with mean follow-up duration of 33.1 mon-ths were analysed in this study (Fig. 1).
[image:4.595.45.544.225.689.2]Among 275 participants, 174 (63%) subjects representing Gro-up I were able to discontinue insulin treatment at least once with a switch to OHAs at a rate of 23% per year. Of these, 122 (44%) subjects in Group IA were treated with insulin therapy for 14.9 months before switching to OHAs and successfully continued with OHAs. Fifty-two (19%) patients in Group IB were treated with insulin therapy for 19.0 months and then with OHAs for 9.2 months before resuming insulin therapy. Group
Fig. 2. HbA1c changes during the study periods. (A) HbA1c changes during the study periods according to groups. (B) HbA1c changes during the study periods between PCGR tertiles at baseline. (C) HbA1c changes during the study periods between PCGR tertiles among Group IA and IB. *p val-ue <0.005, Group IA vs. Group IB, †p value <0.005, Group IA vs. Group II, ‡p value <0.005, PCGR in lowest tertile vs. PCGR in higher tertiles. PCGR,
post-prandial C-peptide-to-glucose ratio.
10.5
10.0
9.5
9.0
8.5
8.0
7.5
7.0
Time (months)
0 5 10 15 20 25 30 35 40 45
HbA1c (%)
PCGR in lowest tertile PCGR in higher tertiles ‡
‡ ‡
‡
B 10.0
9.5
9.0
8.5
8.0
7.5
7.0
6.5
Time (months)
0 5 10 15 20 25 30 35 40 45
HbA1c (%)
Group IA Group IB Group II
*†
*† * †
†
*
A
10.5
10.0
9.5
9.0
8.5
8.0
7.5
7.0
Time (months)
0 5 10 15 20 25 30 35 40 45
HbA1c (%)
PCGR in lowest tertile PCGR in higher tertiles
‡ ‡
II contained 101 (37%) subjects who continued insulin thera-py without modifying the treatment modality.
Clinical and laboratory characteristics of patients at baseline
The demographic and laboratory data of all of the subjects at baseline are shown in Table 1. For all participants, the mean age of patients was 58.5±10.9 years, and the mean duration of diabetes was 9.2 years. The average BMI was 24.5±3.3 kg/m2,
and 45% of patients were defined as obese by the obesity crite-ria of the Asian and Pacific region (BMI ≥25 kg/m2).20,21 BMI was
significantly lower in Group II than in Group I. Group II showed significantly lower BMI than Group I. Of Group I, age, gender, BMI, duration of diabetes, and family history of diabetes were similar between Group IA and Group IB. Regarding glucomet-abolic parameters, HbA1c and postprandial plasma glucose concentration were higher in the subjects who resumed insu-lin treatment (Group IB) than in those who maintained OHAs (Group IA). Levels of PCGR were arbitrarily divided into two subgroups as the lowest tertile and higher tertiles (including the middle and highest tertile). Group IB had a statistically high-er proportion of patients in the lowest thigh-ertile of PCGR than Group IA. Compared to subjects who could be maintained on OHAs (Group IA), subjects who were ultimately treated with insulin therapy (Group IB and Group II) showed a lower BMI (Group IA vs. Group IB+II, 25.1±3.4 kg/m2 vs. 24.2±3.0 kg/m2, p value 0.033), lower proportion of subjects with higher tertiles
of PCGR (76% vs. 59%, p value 0.006), and a lower postprandial SUIT index (39.6±21.7 vs. 32.7±21.8, p value 0.015). There was no statistical difference in the drug regimen before insulin initiation and the incidence of hypoglycemic events between groups. Supplementary Table 1 (only online) presentsd the use of insu-lin and OHAs at baseinsu-line and during the follow-up period.
Changes in glycemic parameters during the follow-up period
Changes in HbA1c during the study period are shown in Fig. 2. All of the groups showed markedly decreased levels of HbA1c by an average of 1.8% (p value <0.001) at a mean insulin dose of 33.9±16.4 U within the first 3 months after insulin initiation (Fig. 2A). During the study period, subjects in Group IA had ad-equate glycemic control with significantly lower levels of HbA1c than subjects in Group IB or Group II. However, there was no significant difference in HbA1c between Group IB and Group II during the period. Fig. 2B shows HbA1c changes according to the baseline PCGR. Compared to subjects with higher ter-tiles of PCGR levels, including the middle and highest tertile, those with the lowest tertile of PCGR level showed higher lev-els of HbA1c from the baseline during the study period. Fig. 2C shows the similar pattern of HbA1c changes among Group IA and IB to those with all subjects in Fig. 2B.
To evaluate glycemic parameters between Group IA and Group IB, the change in HbA1c from the switching point to OHAs until resumption of insulin therapy in Group IB (at 9.2 months after switching to OHAs) and the corresponding mean values in Group IA (at 6–12 months after switching to OHAs) are shown in Fig. 3. Group IA and Group IB showed similar lev-els of HbA1c at the switching point to OHAs from insulin as 7.2±1.3% and 7.4±1.3%, respectively. HbA1c increased by 0.6± 0.04% after the first 3 months of OHAs treatment in Group IB (p value 0.046), whereas Group IA maintained glycemic control at a mean HbA1c level of 7.3±0.1% and without a significant dif-ference during the period. Compared to Group IA, the level of HbA1c in Group IB significantly increased from 8.0±0.3% at 6 months after switching to OHAs to 9.0±0.3% at the time of re-sumption of insulin (p value <0.005).
Predictive parameters for resumption of insulin after initial insulin discontinuation and failure of treatment with OHAs
[image:5.595.58.293.427.670.2]Cox regression analysis was performed to investigate the pre-dictive factors for resumption of insulin therapy for glucose control in subjects who switched to OHAs from initial insulin therapy (Group IA vs. Group IB) (Table 2). We included clini-cally important conventional variables and established pa-rameters that were significantly different between Group IA and Group IB, based on the results of Table 1. Compared to subjects with higher tertiles of PCGR at baseline, subjects with the lowest tertile of PCGR at baseline showed a 3.4-fold in-creased risk of OHA failure and switch to insulin therapy (95% Fig. 3. Changes in HbA1c after switching to oral hypoglycemic agents
(Group IA vs. Group IB). *p value <0.005, Group IA vs. Group IB, †HbA1c
level at the same point of resumption with insulin in Group IB (at 9.2 months after switching to OHAs in Group IB and at 6–12 months after switching to OHAs in Group IA, respectively). OHA, oral hypoglycemic agent.
9.5
9.0
8.5
8.0
7.5
7.0
6.5
Switching to OHAs
3 months 6 months Resumption with insulin
HbA1c (%)
Group IA Group IB
* *
†
CI 1.56–7.31) in Model 1. In the regression model including ad-ditional variables during follow-up period (Model 2), HbA1c levels at 6 months after switching to OHAs [hazard ratio (HR) 1.68, 95% CI 1.01–1.10], insulin dose at switching to OHAs (HR 1.04, 95% CI 1.01–1.07), and the lowest tertile of baseline PCGR (HR 8.34, 95% CI 2.34–29.74) were significant predictors of an increased likelihood of resumption of insulin therapy. Further-more, multivariate linear regression analysis showed that HbA1c level at 6 months after switching to OHAs was significantly in-creased in patients with a lower PCGR level at baseline (stan-dardized β=-0.170, p value 0.038) (Supplementary Table 2, only online).
As shown in Fig. 4, we observed a higher cumulative inci-dence of OHA failure and resumption of insulin in subjects with
the lowest tertile of baseline PCGR than those with higher ter-tiles of baseline PCGR (p value <0.001 by log-rank test).
Predictive parameters for subjects who were ultimately treated with insulin therapy
Multivariate logistic regression analysis was performed to in-vestigate significant factors predicting the likelihood of ultimate inability to maintain adequate glycemic control without using insulin in subjects who are clinically insulin-dependent for glucose control (Group IB and Group II). We found that the low-est tertile of PCGR at baseline (OR 2.22, 95% CI 1.09–4.52) was independently associated with ultimate treatment with insu-lin therapy in subjects in Group IB and II (p value 0.024) (Table 3). In the ROC analysis, 1.490 of PCGR at baseline was the op-timal cut-off value for predicting successful switching to OHAs with sensitivity of 64.5% and specificity of 66.7% (AUC 0.654, p value 0.003) (Supplementary Fig. 1, only online).
DISCUSSION
Considering the advantages of appropriate use of insulin ana-logues to modify the natural history of diabetes through resto-ration and preservation of pancreatic β-cell function,4,5 we
hy-pothesised that T2D patients who have reserved β-cell function might be more likely to be stably maintained with OHAs after switching from initial insulin therapy. To address these ques-tions, we gathered data from three previous cohort studies of insulin treatment and focused on investigating the clinical and metabolic parameters of T2D patients who could maintain adequate glycemic control after switching to OHAs from initial insulin therapy. From this longitudinal study, we found that 63% of enrolled participants (Group I) were able to stop insulin therapy and switch to OHAs at a rate of 23% per year. Among these patients, 44% (Group IA) could maintain treatment with OHAs and 19% (Group IB) required resumption of insulin af-Table 2. Cox Regression Analysis of Resumption with Insulin after Insulin Discontinuation and OHA Therapy (Group IA, n=122; Group IB, n=52)
Variables Model 1 Model 2
HR (95% CI) p value HR (95% CI) p value
Age (yr) 1.00 (0.96–1.03) 0.757 1.01 (0.96–1.06) 0.776 Sex (women=0, men=1) 2.39 (1.20–4.77) 0.013 1.83 (0.58–5.79) 0.303 BMI (kg/m2) 0.98 (0.88–1.08) 0.662 1.00 (0.83–1.20) 0.984
Duration of diabetes (yr) 1.02 (0.96–1.08) 0.586 1.02 (0.92–1.13) 0.759 Fasting plasma glucose (mg/dL) 1.00 (0.99–1.00) 0.488 0.99 (0.98–1.00) 0.057 HbA1c (%) 1.02 (0.81–1.29) 0.846 1.02 (0.71–1.47) 0.927 PCGR
Higher tertiles Referent Referent
[image:6.595.43.541.85.252.2]Lowest tertile 3.38 (1.56–7.31) 0.002 8.34 (2.34–29.74) 0.001 Insulin dose at the time switching to OHAs (U) - - 1.04 (1.01–1.07) 0.033 HbA1c at 6 months after switching to OHAs (%) - - 1.68 (1.01–1.10) 0.015 BMI, body mass index; CI, confidence interval; HR, hazard ratio; OHA, oral hypoglycemic agent; PCGR, postprandial C-peptide-to-glucose ratio.
Fig. 4. Kaplan-Meier curves for events of resumption with insulin ac-cording to the tertiles of postprandial C-peptide-to-glucose ratio at base-line. PCGR, postprandial C-peptide-to-glucose ratio.
100
80
60
40
20
0
0 10 20 30 40 50 60 Time (months)
p value <0.001 by log-rank test
Events (%)
[image:6.595.45.279.289.515.2]ter OHA failure. However, 37% of the enrolled subjects (Group II) were unable to discontinue insulin at any time during the study observation.
Investigation of the predictive factors for requiring insulin therapy to maintain optimal glucose control yielded two main findings in Korean T2D patients: 1) a high HbA1c level at 6 months after switching to OHAs was the most significant inde-pendent parameter associated with inadequate glycemic con-trol upon switching to OHAs from insulin analogues, indicating its importance in guiding therapeutic strategy for maintaining an optimal glycemic target; 2) the level of PCGR, which reflects preserved dynamic β-cell function22,23 before initiation of
insu-lin analogues could be a predictive parameter of whether pa-tients could be switched to OHAs or should be maintained with insulin analogues.
To appropriately evaluate predictive independent factors, we adopted parameters reflecting β-cell function indices and the time-to event relationship of therapeutic changes using the Cox regression model. In numerous previous studies, secreto-ry pancreatic β-cell dysfunction has been regarded as an im-portant factor for glycemic control, especially in Asian popu-lations,24 whereas C-peptide measurement is a well-known
method for assessing β-cell function.25 Recent studies
suggest-ed that the C-peptide-to-glucose ratio might be a better mark-er to evaluate endogenous insulin secretory ability, especially in the postprandial state, because elevated levels of plasma glucose and incretin hormone further stimulate insulin secre-tion.26-28 Also, postprandial insulin deficiency is considered the
main determinant of deteriorating glycemic control in patients with newly detected T2D.29 Previously, Saisho, et al.30 reported
PCGR as a predictor of subsequent insulin treatment in patients with T2D, showing a similar clinical implication of PCGR to the findings of the present study. While conventional clinical variables such as duration of diabetes and type of OHA showed no clinical implication in predicting the need for insulin con-tinuation to maintain optimal glucose control at the time of
in-sulin initiation, a lower baseline PCGR was found to be a strong independent variable for failure of OHAs and ultimate re-quirement for insulin therapy after switching to OHAs (Table 2 and 3, Fig. 4). Regarding changes in HbA1c according to ther-apeutic strategies, HbA1c levels at 3 to 6 months after chang-ing to OHA were significantly different between Group IA sub-jects who continued OHAs and Group IB subsub-jects who failed on OHAs and resumed insulin therapy (Fig. 3). Furthermore, in contrast to Group IA, inadequate glycemic control was ob-served in both Group IB and II, who ultimately required insu-lin therapy, with no significant differences in HbA1c level be-tween these two groups (Fig. 2A). However, Cox regression analysis showed that HbA1c at 6 months after switching from insulin to OHA was the only significant predictor for resump-tion of insulin after OHA failure among the follow-up param-eters tested. In multivariate linear regression analysis, we addi-tionally revealed a negative correlation between baseline PCGR and HbA1c at 6 months after switching from insulin therapy to OHAs (Table 3). With respect to prediction of the best ther-apeutic strategy for T2D patients after insulin initiation, these data indicate that glycemic status after switching from insulin therapy to OHAs is affected by the baseline secretory β-cell function.
It is known that intensive insulin therapy in patients with newly diagnosed T2D with severe hyperglycemia could in-duce favourable long-term glycemic control with recovery or a decreased rate of deterioration of β-cell function.4,5,31
Al-though the Korean T2D patients enrolled in this study were not newly diagnosed but had an average disease duration of 9.2 years, subjects in Group IA showed a statistically signifi-cant improvement in the level of PCGR (increased by 0.723± 1.717, p value 0.001) and lipid profiles (data not shown), to-gether with a significant decrease in HbA1c, with insulin ther-apy. This finding might imply that the implementation of insu-lin would result not only in the possibility of adequate glycemic maintenance with a switch to OHAs, but also restoration and preservation of pancreatic β-cell function even in subjects with a moderate duration of diabetes who have relatively pre-served insulin secretory function assessed by PCGR. There-fore, PCGR at baseline might be the strongest candidate for predicting sustained glycemic control with OHAs among sub-jects who change from insulin therapy. Moreover, PCGR after glycemic control with insulin therapy could have potentials to expect the glucose control after the discontinuation of insulin therapy. The possible explanation for this finding might be the restoration of the lost secretory function of endogenous insu-lin after exogenous insuinsu-lin therapy. Therefore, further research will be warranted for investigating the ways to restore the lost insulin function. As the duration of diabetes increases, insulin secretion decreases, especially postprandial secretion.29
[image:7.595.56.295.105.235.2]Con-sidering the natural course of T2D, early initiation of insulin administration in subjects with escalating hyperglycemia could be a preferable option to restore and maintain pancreatic β-cell Table 3. Multivariate Logistic Regression Models for the Variables
Independently Associated with Subjects Who Ultimately Treated with Insulin: 0=Group IA (n=122), 1=Group IB and II (n=153)
Variables Odds ratio (95% CI) p value
Age (yr) 0.24 (0.96–1.01) 0.235 Sex (women=0, men=1) 1.54 (0.85–2.78) 0.154 BMI (kg/m2) 0.93 (0.84–1.02) 0.106
Duration of diabetes (yr)* 1.77 (0.71–4.41) 0.217 Fasting glucose (mg/dL) 1.00 (0.99–1.00) 0.517 HbA1c (%) 1.17 (0.92–1.49) 0.211 PCGR
Higher tertiles Referent
Lowest tertile 2.22 (1.09–4.52) 0.029 BMI, body mass index; CI, confidence interval; PCGR, postprandial C-peptide-to-glucose ratio.
function and provide better glycemic control.
The lowest tertile of the postprandial SUIT index in this study presented a significantly higher HR for predicting resumption with insulin after insulin discontinuation and OHA therapy (HR 2.80, 95% CI 1.29–6.04, p value 0.009) (Supplementary Table 3, only online), but that of the fasting SUIT index did not (HR 1.05, 95% CI 0.48–2.29, p value 0.898). In contrast to Ya-mada, et al.’s17 report of significant correlation between fasting
SUIT index and stimulated serum C-peptide (p value <0.001) and our present result of significant correlation between fast-ing SUIT index and PCGR level (standardized β=0.409, p value <0.0001), the postprandial SUIT index was a stronger predic-tor of an increased likelihood of resumption of insulin therapy relative to fasting measures in the present study. Since we could not assess the AUC120 min SUIT index, further in-depth analysis
of the augmentation of C-peptide in response to the increase of glucose during the oral glucose tolerance test (OGTT), which has previously been shown to indicate β-cell function well19 is
also warranted.
There are some limitations in this study that should be ad-dressed by further trials. First, despite the existence of baseline data from the longitudinal study by recruiting three insulin co-hort studies, the retrospective design of this study has inherent limitations. However, we focused on the experience in the set-ting of real-world practice. Second, data included in this study did not cover the patients’ satisfaction with the drugs and any drug-related side effects such as weight gain. Third, during the study period, the decision of when to switch to OHAs and re-sume insulin therapy was made at the discretion of the patients’ physicians. However, the levels of HbA1c at switching to OHAs from initial insulin therapy in Group IA and Group IB turned out similarly to be under 7.5%.
In conclusion, a higher level of PCGR at the time of insulin initiation and lower HbA1c level at 6 months after switching to OHAs are clinical parameters that should be taken into consid-eration as important predictors for the successful maintenance of OHAs after switching from insulin therapy in Korean patients with T2D.
REFERENCES
1. DeFronzo RA. Lilly lecture 1987. The triumvirate: beta-cell, muscle, liver. A collusion responsible for NIDDM. Diabetes 1988;37:667-87. 2. Kahn SE. The relative contributions of insulin resistance and be-ta-cell dysfunction to the pathophysiology of type 2 diabetes. Dia-betologia 2003;46:3-19.
3. Kim YJ, Jeon JY, Han SJ, Kim HJ, Lee KW, Kim DJ. Effect of socio-economic status on the prevalence of diabetes. Yonsei Med J 2015; 56:641-7.
4. Weng J, Li Y, Xu W, Shi L, Zhang Q, Zhu D, et al. Effect of intensive insulin therapy on beta-cell function and glycaemic control in patients with newly diagnosed type 2 diabetes: a multicentre ran-domised parallel-group trial. Lancet 2008;371:1753-60.
5. Kramer CK, Zinman B, Retnakaran R. Short-term intensive insu-lin therapy in type 2 diabetes mellitus: a systematic review and
meta-analysis. Lancet Diabetes Endocrinol 2013;1:28-34. 6. Tibaldi J, Rakel RE. Why, when and how to initiate insulin therapy
in patients with type 2 diabetes. Int J Clin Pract 2007;61:633-44. 7. Yoo HJ, Park KY, Park KS, Ahn KJ, Min KW, Park JH, et al. Safety and
efficacy of modern insulin analogues. Diabetes Metab J 2013;37: 181-9.
8. Riddle MC, Rosenstock J, Gerich J; Insulin Glargine 4002 Study In-vestigators. The treat-to-target trial: randomized addition of glargine or human NPH insulin to oral therapy of type 2 diabetic patients. Diabetes Care 2003;26:3080-6.
9. Holman RR, Farmer AJ, Davies MJ, Levy JC, Darbyshire JL, Keenan JF, et al. Three-year efficacy of complex insulin regimens in type 2 diabetes. N Engl J Med 2009;361:1736-47.
10. Janka HU, Plewe G, Riddle MC, Kliebe-Frisch C, Schweitzer MA, Yki-Järvinen H. Comparison of basal insulin added to oral agents versus twice-daily premixed insulin as initial insulin therapy for type 2 diabetes. Diabetes Care 2005;28:254-9.
11. Hwang YC, Kang JG, Ahn KJ, Cha BS, Ihm SH, Lee S, et al. The gly-cemic efficacies of insulin analogue regimens according to base-line glycemic status in Korean patients with type 2 diabetes: sub-analysis from the A(1)chieve(®) study. Int J Clin Pract 2014;68: 1338-44.
12. Lee YH, Lee BW, Chun SW, Cha BS, Lee HC. Predictive character-istics of patients achieving glycaemic control with insulin after sulfonylurea failure. Int J Clin Pract 2011;65:1076-84.
13. Choe EY, Lee YH, Lee BW, Kang ES, Cha BS, Lee HC. Glycemic ef-fects of once-a-day rapid-acting insulin analogue addition on a basal insulin analogue in Korean subjects with poorly controlled type 2 diabetes mellitus. Diabetes Metab J 2012;36:230-6. 14. Lee YH, Lee BW, Kwon HJ, Kang ES, Cha BS, Lee HC. Higher
morn-ing to evenmorn-ing ratio in total dose of twice-daily biphasic insulin ana-log might be effective in achieving glucose control in patients with poorly controlled type 2 diabetes. Diabetes Technol Ther 2012; 14:508-14.
15. Jung CH, Park JY, Cho JH, Yoon KH, Yang HK, Lee YH, et al. The optimal morning:evening ratio in total dose of twice-daily bipha-sic insulin analogue in poorly controlled Type 2 diabetes: a 24-week multi-centre prospective, randomized controlled, open-la-belled clinical study. Diabet Med 2014;31:68-75.
16. Kim W, Kim KJ, Lee BW, Kang ES, Cha BS, Lee HC. The glycated albumin to glycated hemoglobin ratio might not be associated with carotid atherosclerosis in patients with type 1 diabetes. Dia-betes Metab J 2014;38:456-63.
17. Yamada Y, Fukuda K, Fujimoto S, Hosokawa M, Tsukiyama K, Na-gashima K, et al. SUIT, secretory units of islets in transplantation: An index for therapeutic management of islet transplanted patients and its application to type 2 diabetes. Diabetes Res Clin Pract 2006;74:222-6.
18. Lee EY, Hwang S, Lee SH, Lee YH, Choi AR, Lee Y, et al. Postpran-dial C-peptide to glucose ratio as a predictor of β-cell function and its usefulness for staged management of type 2 diabetes. J Di-abetes Investig 2014;5:517-24.
19. Ito R, Fukui T, Hayashi T, Osamura A, Ohara M, Hara N, et al. Teneligliptin, a dipeptidyl peptidase-4 inhibitor, improves early-phase insulin secretion in drug-naïve patients with type 2 diabe-tes. Drugs R D 2015;15:245-51.
20. Park SY, Cho YJ, Lee SR, Chung H, Jeong K. Triglyceride is a useful surrogate marker for insulin resistance in Korean women with polycystic ovary syndrome. Yonsei Med J 2015;56:785-92. 21. World Health Organization, Regional Office for the Western
Pa-cific. The Asia Pacific perspective: redefining obesity and its treat-ment. Sydney: Health Communications Australia; 2000.
et al. Evaluation of insulin sensitivity and beta-cell function in-dexes obtained from minimal model analysis of a meal tolerance test. Diabetes 2004;53:1201-7.
23. Mari A, Camastra S, Toschi E, Giancaterini A, Gastaldelli A, Min-grone G, et al. A model for glucose control of insulin secretion dur-ing 24 h of free livdur-ing. Diabetes 2001;50 Suppl 1:S164-8.
24. Kim DJ, Lee MS, Kim KW, Lee MK. Insulin secretory dysfunction and insulin resistance in the pathogenesis of Korean type 2 diabe-tes mellitus. Metabolism 2001;50:590-3.
25. Törn C. C-peptide and autoimmune markers in diabetes. Clin Lab 2003;49:1-10.
26. Meier JJ, Menge BA, Breuer TG, Müller CA, Tannapfel A, Uhl W, et al. Functional assessment of pancreatic beta-cell area in humans. Diabetes 2009;58:1595-603.
27. Drucker DJ, Nauck MA. The incretin system: glucagon-like
pep-tide-1 receptor agonists and dipeptidyl peptidase-4 inhibitors in type 2 diabetes. Lancet 2006;368:1696-705.
28. Fleetcroft R, Schofield P, Ashworth M. Variations in statin prescrib-ing for primary cardiovascular disease prevention: cross-sectional analysis. BMC Health Serv Res 2014;14:414.
29. Shim WS, Kim SK, Kim HJ, Kang ES, Ahn CW, Lim SK, et al. Dec-rement of postprandial insulin secretion determines the progres-sive nature of type-2 diabetes. Eur J Endocrinol 2006;155:615-22. 30. Saisho Y, Kou K, Tanaka K, Abe T, Kurosawa H, Shimada A, et al.
Postprandial serum C-peptide to plasma glucose ratio as a predic-tor of subsequent insulin treatment in patients with type 2 diabetes. Endocr J 2011;58:315-22.
